Acouple of articles ago I wrote about when we should normally expect winter to start, and I added a little bit of information about the probability of receiving different amounts of snow in a single day. Well, it seems my snowfall probability comments have prompted a fair bit of discussion, as I have received a fair number of emails about this. So in response this week I’ll take a deeper look into snowfall.
Years ago, during one of my university courses on water systems, we were looking at stream flow and how to calculate the probability of 25-, 50-and 100-year floods. Once I learned how to do this, I started thinking about snowstorms and how often we should expect big storms to hit us. So, I went through all the data for Winnipeg (a nice central region) and applied the same probability techniques I learned for stream flooding, since the statistical theory is sound for both things. Well, good thing I kept all that data, so let’s dig into the numbers and see what we come up with.
First of all, I think I have to point out that when we look at snowfall data there are different periods in which you can measure snowfall, or precipitation for that matter. First of all, you can measure the amount of snow that falls in one day, which would be from midnight to midnight. This one is easy enough since this is the way precipitation amounts are recorded. We all know that Mother Nature doesn’t have a very accurate clock, so we can’t always count on a snowfall occurring nicely during this one-day period. Another way to measure snowfall is to calculate how much fell in a given 24-hour period. To calculate this, you need access to hourly data which, unfortunately, is not readily available. But we can measure snowfall a third way which does help us “catch” these 24-hour-period snowfalls, and that’s to look at what I call asnow event.A snow event is when we record snowfall amounts on two or more consecutive days.
Let’s begin our look by examining snowfall days. In Winnipeg, during any given winter, there is a 90 per cent chance that there will be about 30 days when snow falls. About 50 per cent of the time we will see around 45 snowfall days and there is less than a one-in- 100 chance of seeing more than 70 snowfall days in any given winter. Keep in mind that a snowfall day is any day that records snow, no matter how small. Now, when we look at how much snow we can expect to fall during a snowfall day we find that 90 per cent of the time we can expect 0.2 centimetres of snow or more. About 50 per cent of the time we can expect more than two cm of snow, which means about half the time it snows we see less than two cm of snow! If we look at how often we should expect five cm or more, we would find that the probability has fallen to about 10 per cent. This means that if we get 30 snowfall days during the winter, only about three of them will give us more than five cm of snow. Looking at higher one-day snowfall amounts, we find an amount of 10 cm or more occurs about once in every 25 snowfall days, or once, maybe twice, per winter. Getting more than 30 cm of snow in one day is extremely rare. The probability of this occurring is about 0.1 per cent, which means about one in every 1,000 snowfall days, or around once every 20 to 40 years, we could expect to see this much snow in a single day!
Now, as I pointed out earlier, Mother Nature rarely dumps all the snow conveniently during a midnight-to-midnight 24-hour period, so let’s look at snowfall events and see how the numbers shape up.
In any given winter we typically see about 20 snowfall events where we record snowfall on two or more consecutive days. About half the time we could see as many as 28 snowfall events, and once in every 100 winters we could experience as many as 40 snowfall events. Just like with the one-day snowfall amounts, the majority of these snowfall events only record small amounts of snow. About 90 per cent of the time we would expect to see 0.5 cm or more snowfall and about 50 per cent of the time we would expect to see more than two cm. Interestingly, these numbers are just about the same as for the one-day snowfall probabilities. If we look at the probability of receiving more than five cm of snow during a snowfall event we would see that it happens about 30 per cent of the time. Going up to 10 cm or more, our probability drops to about 10 per cent or around twice per winter. How often should we expect to see big storms with more than 30 cm of snow? Well, according to my probabilities, it would be about one in every 200 snowfall events or about once every 10 years.
Well, these numbers sound like they might be off, so let’s take a look at last winter. Last winter I measured two days with more than five cm of snow; four snow events with more than five cm; and only one snow event with more than 10 cm. This was fairly close to what was expected, according to the probabilities. You do have to remember that probabilities are just that, probabilities. This means we could see much more or much less than what these numbers indicate in any one winter.